Managing conflict is one of the tasks PMI lists under the People domain, and the examination tests it through scenarios in which two people on a project disagree and the candidate must choose what the project manager does next. The material behind those questions is a set of five responses to conflict, taken from the work of Kenneth Thomas and Ralph Kilmann in the 1970s, and a distinction that comes before any of them, which is what the conflict is about.
Conflict on a project is not one thing. Disagreement about the work is a group doing its job, because two people who both understand the problem and reach different conclusions have produced information the project needed. Disagreement about people is different in kind, because there is nothing to resolve and nothing to learn from it. The first is worth having and the second is worth stopping, and the practical skill is telling them apart quickly, usually before either party has.
The tell is what a question about the substance produces. Asked why an approach was chosen, a person in a task conflict gives an answer about the approach. A person in a relationship conflict gives an account of the other person.
Where project conflict comes from
Most conflict on a project is structural. Schedules, priorities and resources produce more of it than personalities do, because the organisation has committed the same people to two things at once or funded a scope it will not staff. A recurring argument between two leads who are measured on incompatible targets looks like a personality clash and is nothing of the kind, and a project manager who sets out to mediate it is trying to fix with a conversation something no conversation can fix. The remedy for a structural conflict is a change to the structure, which is usually a decision somebody more senior has been avoiding.
The five conflict modes
Where the conflict is genuinely between two people, five responses are available, and none of them outranks the others. Each is a setting on two dials. The first dial is how hard a person presses for their own outcome, and the second is how much they work for the other side's.
Assertiveness is how hard a person presses for their own outcome, and cooperativeness is how much work they do for the other side's. Compromise sits halfway along both, which is what makes it quick and lossy. Collaborating is the only mode high on both, and the only one that can beat both opening positions.
| Mode | What the person does | One situation where it is genuinely the right choice |
|---|---|---|
| Withdraw or avoid | Steps back and leaves the issue unresolved | Two leads are shouting an hour before a steering committee, and the same conversation on Thursday will cost a tenth as much |
| Smooth or accommodate | Concedes their position to preserve the relationship | The supplier is right about the interface and the project manager is wrong, and conceding early buys credit for the argument that is coming |
| Compromise or reconcile | Each side gives something up to reach a settlement | Two teams want the same integration window, the claims are of equal weight and the release is on Friday |
| Force or direct | One view is imposed, usually by whoever holds authority | A proposed change would breach a regulatory constraint, or a deadlock has already cost a week of everybody's time |
| Collaborate or problem solve | Works the interests underneath the positions until a solution both sides want appears | Two departments disagree about a reporting design they will both live with for years, and there is a fortnight to get it right |
What each mode costs
Withdrawing costs nothing when the issue will resolve itself and a great deal when it will not. Avoiding a problem that will still be there, and larger, in a month is a different act that looks the same, and the two are told apart by asking what changes if nobody touches it.
Smoothing spends the person's own position to buy goodwill, which works as a move and fails as a habit. A person who concedes everything stops being worth consulting, because their agreement carries no information about whether the idea was any good.
Compromise is genuinely lossy, since each side gives something up and neither gets what it wanted. That is a fair price between two positions of roughly equal weight under time pressure, where a workable settlement beats a perfect one. It is a poor price on a safety or regulatory constraint, where it produces a decision neither side would defend afterwards.
Forcing is the right answer more often than people like to admit, and it spends authority every time it is used. Safety, legal exposure, a live incident and a deadlock that has already cost a week all justify somebody simply deciding. The losing side has been overruled and has not been persuaded, so forcing works as an exception and corrodes a team as a routine.
Collaborating is the only mode that can produce an outcome better than either opening position, because it goes underneath the positions to the interests holding them up. What it costs is time, and whether that price is worth paying is the next question.
The limits of collaborating
Collaborating is the mode most often rewarded in scenario questions, which is reasonable, since it is the one that produces durable agreements and the one inexperienced managers reach for least. Treating it as always correct is still a mistake.
It is expensive. It needs time both parties actually have, a relationship strong enough to survive the digging and an issue with enough at stake to be worth the effort. Spending an afternoon collaborating over which room a workshop uses is a failure to gauge the size of a problem, and collaborating during an outage means nobody is fixing the outage.
The honest formulation is that collaborating is the default and the other four are choices made against it for a stated reason. A project manager who can say why they forced this one is doing what the modes exist for.
Escalation and its threshold
Sometimes none of the five settles the matter, and the remaining option is escalation. Escalation is a last resort because it moves the decision away from the people who understand it and spends the goodwill of whoever receives it. It is still a legitimate move, and holding a conflict past the point where it could have been settled locally is worse than escalating it.
A usable threshold has three parts. The disagreement has to be blocking work, both parties have to have tried in good faith and be willing to say so, and the decision has to sit outside the authority of everyone in the room. When all three hold, the project manager escalates the same day, carries both positions and tells the other party first.